Re: [AD] windows resizing regression

[Peter Wang <novalazy@xxxxxxxxxx>]

On 2010-11-08, Evert Glebbeek <eglebbk@xxxxxxxxxx> wrote:
> On 8 Nov 2010, at 22:41 , Trent Gamblin wrote:
> > Just look at the examples. ex_resize, ex_fs_resize, and ex_fs_window are the only
> > ones that call al_resize_display. And ex_fs_window is the only one that currently
> > doesn't have a call to al_acknowledge_resize. It was planned that way from the beginning by Elias and I. Something may have changed that I missed, or something may
> > need to change, and that's fine...
> 
> What I described is what I've always understood the behaviour to be.

It has been so at least since r8025.

    Author: Elias Pschernig
    Date:   Sat Jun 30 16:18:47 2007 +0000

    Implemented immediate-resize for al_resize_display, and implemented fullscreen resizing.

> 
> > And the original reason for the al_acknowledge_resize was that on X11 you have to wait
> > for an event to come to tell you when you can resize the display, which could take an
> > indeterminate amount of time.
> 
> I thought it was about not changing the size of the backbuffer on the user in the middle of a graphic update.

Yes.  It goes back at least as early as Dec 2006, when Elias writes on [AD]:

    The problem is, doing the resizing asynchronously, that is, change
    internal display dimensions, clipping, OpenGL setup, and so on,
    asynchronously from the background thread whenever an X11 resize event
    is received, seems like a very bad idea.

al_acknowledge_resize is introduced to react to the user resizing the
display.

Peter